Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Transferring Experience from Simulation to the Real World for Precise Pick-And-Place Tasks in Highly Cluttered Scenes

: Kleeberger, Kilian; Völk, Markus; Moosmann, Marius; Thiessenhusen, Erik; Roth, Florian; Bormann, Richard; Huber, Marco

Fulltext (PDF; )

Institute of Electrical and Electronics Engineers -IEEE-; IEEE Robotics and Automation Society; IEEE Industrial Electronics Society; International Robotics Society of Japan; New Technology Foundation; Society of Instrument and Control Engineers of Japan:
IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2020 : Consumer Robotics and Our Future. October 25 - October 29, 2020, Las Vegas, NV, USA, On-Demand Conference, Virtual
Piscataway, NJ: IEEE, 2020
ISBN: 978-1-7281-6211-9
International Conference on Intelligent Robots and Systems (IROS) <2020, Online>
Conference Paper, Electronic Publication
Fraunhofer IPA ()
bin-picking; maschinelles Sehen; maschinelles Lernen

In this paper, we introduce a novel learning-based approach for grasping known rigid objects in highly cluttered scenes and precisely placing them based on depth images. Our Placement Quality Network (PQ-Net) estimates the object pose and the quality for each automatically generated grasp pose for multiple objects simultaneously at 92 fps in a single forward pass of a neural network. All grasping and placement trials are executed in a physics simulation and the gained experience is transferred to the real world using domain randomization. We demonstrate that our policy successfully transfers to the real world. PQ-Net outperforms other model-free approaches in terms of grasping success rate and automatically scales to new objects of arbitrary symmetry without any human intervention.